Sex ratios in ant colonies are widely variable, an observation previously unresolyed by theories of selection acting on populations to reach an equilibrium point. A model based on n-person game theory, in which colonies are assumed to share information on reproduction, shows that there is no optimal sex ratio, particularly for offspring of mound-building ant species, because assumptions allowing the existence of a core to the game are violated. Data on the reproductive outputs from colonies of Formica obscuripes in Illinois illustrate the model's features. Although some trends in sex-ratio change among colonies were observed in consecutive years, the within-nest variability in male frequency remained too large to describe the system as operating near equilibrium. Reports of other ant species indicate similar levels of variability in sex ratio between nests. That no optimum exists, as predicted from game theory, is a likely explanation for the confusion of progeny sex ratios in Hymenopteran societies.